2 Quality Assurance vs. Quality Control Quality Control - QC refers to the measures that must be included during each assay run to verify that the test is working properly.Quality Assurance - is concerned with much more: that the right test is carried out on the right specimen, and that the right result and right interpretation is delivered to the right person at the right time.Quality AssuranceQuality ControlMuch more: An overallmanagement plan toguarantee theintegrity of data(The “system”)A series ofanalyticalmeasurements to ensure that the results generated by the test system are correct
4 Definition - QC in the medical laboratory QC-Statistical process - to monitor & evaluate analytical process which produces patient results to ensure that medical decisions can be taken with confidenceEnsures continuous monitoring of performance of test analytesQC result - may be quantitative, qualitative (positive or negative) or semi- quantitative (limited to a few different values)A total QC system must control both trueness and precisionIQC and EQA (External Quality Assessment) are complementary in ensuring the reliability of test results
6 ? True Value - UnknownTrue value - this is an ideal concept which cannot be achieved.Accepted true value - the value approximating the true value, the difference between the two values is negligible or within acceptable limitsError - the discrepancy between the result of measurand and the true (or accepted true value).
7 Accuracy Vs. Precision Accuracy How well a measurement agrees with an accepted valuePrecisionHow well a series of measurements agree with each other
9 Precise & AccurateBias and imprecision are most important at the medical decision levels!For example, ß-hCG clinical decision levels at low concentrations (corresponding to early pregnancy in the female and early testicular cancer in the male) or at moderate concentrations (to diagnose the progression of pregnancy)
10 How to carry out analysis of data? Need tools for data management and analysisBasic statistics skillsManual methodsArithmetic Graph paperCalculatorComputer helpfulMS Excel SoftwareSpreadsheetQC Data Management Software programmesBiorad Unity Realtime, Biorad DesktopRandox 247Medlab QCAutomated Analyzer softwareImportant skills for laboratory personnel
11 Establishing Control Ranges Select appropriate controls & use product insert ranges only as guidelinesManufacturer Ranges are based on reagent lots and materials available at the time of value assignment. During the life of the control lot, manufacturers may reformulate tests or begin using a new source of raw materials for kit/reagent production.Published ranges cannot account for variables such as instrumentation software updates or performance differences over timeCalculate Lab defined Mean & SD:Provisional Mean - Collect database with min 20 data points from separate analytical runsPurpose - covers day to day sources of variability in the measurement procedure to be reasonably represented in the mean value due to calibration frequency, change of reagent or reagent lot, operator technique, temperature/humidity of testing location, daily/weekly maintenance, etc.If the desired 20 data points from 20 separate analytical runs are not available, provisional values may have to be established from data collected over fewer than 20 daysFour control measurements per day for five different days
12 A certain amount of variability will naturally occur when a control is tested repeatedly The goal is to differentiate between variability due to chance than due to errorDetermine the degree of variability in the data to establish acceptable rangeCalculate mean, standard deviation, coefficient of variation; determine target rangesPlot results to develop Levey-Jennings charts- allows you to visually review data points plotted against a ±3SD rangeThe initial assessment of imprecision may not include measurement variability due to influence of factors that occur over a longer time period such as recalibration, reagent and calibrator lot changes, instrument maintenance, and environment variablesTherefore, this provisional mean and range can be followed till 90 days. Then recalculate lab mean and range over 90 days period to include all possible variations and apply till the end of lot.New lot of control material is analyzed for each analyte of interest in parallel with the lot of control material in current use
13 Alternate QC - Tests for which calibration/control material is not available, following alternate quality control measures can be applied:Retesting of any randomly chosen retained samples normal or abnormalRetained proficiency testing material/reference materialSample from healthy volunteer or staff known to be free from any diseaseReplicate test of sample by different method, different machine and different person, wherever applicable
14 Calculation of MeanData set (30.0, 32.0, 31.5, 33.5, 32.0, 33.0, 29.0,29.5, 31.0, 32.5, 34.5, 33.5, 31.5, 30.5, 30.0, 34.0,32.0, 32.0, 35.0, 32.5.) mg/dLThe sum of the values (X1 + X2 + X3 … X20) divided by the number (n) of observationsThe mean of these 20 observations is (639.5 20) = 32.0 mg/dL
15 Measures of Dispersion or Variability There are several terms that describe the dispersion or variability of the data around the mean:RangeVarianceStandard DeviationCoefficient of Variation
16 It is the simplest measure of dispersion RangeRange is the difference or spread between the highest and lowest observationsIt is the simplest measure of dispersionIt makes no assumption about the central tendency of the data
17 Normal DistributionAll values are symmetrically distributed around the meanCharacteristic “bell-shaped” curveAssumed for all quality control statistics
19 Standard Deviation and Probability In general, laboratories use the +/- 2 SD criteria for the limits of the acceptable range for a testWhen the QC measurement falls within that range, there is 95.5% confidence that the measurement is correctOnly 4.5% of the time will a value fall outside of that range due to chance; more likely it will be due to errorTake CAs
20 Calculation of Standard Deviation SD - The standard deviation measures a test's precision or how close individual measurements are to each other.The standard deviation (SD) is the square root of variance or average squared deviation from the meanSD is commonly used due to the same units as the mean and the original observationsSD is the principle calculation used to measure dispersion of results around a meanA high standard deviation can be attributed to:Inherent variability in the test, which represents expected errorAnalytical system malfunction, which represents unexpected error that the laboratory must investigate and correct
21 Coefficient of Variation The Coefficient of Variation (CV) is the standard Deviation (SD) expressed as a percentage of the mean-Also known as Relative Standard deviation (RSD)CV % = SD x 100Mean
22 Precision: CV% Vs. SDThe CV is more accurate comparator than SD because latter typically increases as the conc. of analyte increasesA laboratorian can be easily misled if he is comparing precision for two different methods (e.g. by instrument, method, reagent, etc.) by using standard deviationHave a look:SD of hexokinase and glucose oxidase is 4.8 and 4.0 respectively. Based on just SD, one might conclude that the glucose oxidase method is more precise than the hexokinase methodHowever, CV % for both methods is 4% which shows that the methods are equally precise. Assume the mean for the hexokinase method is 120 and the glucose oxidase mean is 100.
25 lj with Manufacturer & Lab Calculated Mean & SD
26 Quality Improvement : Monthly CV% Monitoring Target CV% (Manufacturer) ASSAYRicos CV%Target CV% (Manufacturer)JanFebMarAprMayRemarksLot No.GLUCOSE220.127.116.11.84.365.592.61OKCRTN18.104.22.168.39.416.568.69BADUREA6.21.564.742.45.314.542.66CHOLESTEROL3.845.864.13TRIGLYCERIDE10.51.775.613.433.861.88HDL3.63.1125.38.873.762.87ObserveSGPT12.22.369.074.35.237.82.5SGOT61.667.935.756.811.14ALKALINE3.21.614.9236.194.15URIC ACIDNA1.943.871.41.513.052.97Remarks: Creatinine review and take corrective actions, validate if reqd
27 AccuracyBias- measures how far your observed value is from a target value (peer group/consensus value)Bias = Lab Value – True Value ; eg Bias is 10 if obtained value for Glucose is 110 & target is 100Bias% = 10% for GlucoseSDI: Standard Deviation Index is a measurement of expressed as increments of SDSDI = Your Mean - Peer Group Mean ……… ? how close your value is to the target valuePeer Group SDSDI = perfect - indicates no difference between lab mean and the consensus group meanZ-score is the no. of standard deviations a control result is from the expected meanEg. Z-score of 2.3 observed value is 2.3 SD away from the expected mean
28 Total Error and TEa TE lab =Labbias (%) + 2.33 CVLab ……………………p<0/01 Concepts of total error (TE) and total allowable error (TEa) - useful to choose SPC rulesTotal error (TE) = Bias% + CV% (Imprecision)TE lab =Labbias (%) CVLab ……………………p<0/01Six Sigma- useful for quantifying test performance- In an ideal world, all our processes would be six sigma, and we could monitor them with very simple QCSigma = TEa - Bias / CVDetermine Quality Requirements for the Test:Biological variation informationClinicians' opinions, National and international expert bodies and agencies, Expert local groups or individualsPublished professional recommendations eg Scandinavian Journal of Clinical and Laboratory Investigation 1999; 59: 585Performance goals set by: Regulatory bodies and agencies, Organizers of External Quality Assessment (EQA) schemesGoals based on the current state of the art, including Inter-laboratory comparison programs, EQA or Proficiency Testing schemes
29 Frequency of QC runGLP - test normal and abnormal controls at least daily when patient testing is performed. If the test is stable for less than 24 hours or some change has occurred which could potentially affect the test stability, controls should be assayed more frequently.Eg, a diabetic patient in a critical care situation may have glucose levels run every 2 to 4 hours. In this case, it is important for the glucose test to be precise because lack of precision can cause loss of test reliability. If there is a lot of variability in the test performance (high imprecision, high standard deviation), the glucose result at different times may not be true.
30 NABL 112 Guidelines W/L is <25/day --- 1 Level of QC W/L is >25/day Levels of QCW/L is >75/day Levels of QC -- twice a day at definite intervals*Min: at least one level of QC whenever patient samples are testedQuality Controls Preparation, Use and QC Protocols:As per manufacturer instructions in Kit insertStorage and Stability of ControlsList of Parameters to be Analyzed with schedule of runs
31 Internal QC Warning 12SD or 1-2s: It is violated if the single IQC Multi control QC rules (WESTGARD RULES) given by Dr. James Westgard of the University of Wisconsin in an article in 1981 on laboratory quality control that set the basis for evaluating analytical run quality for medical laboratories.The Westgard system -based on the principles of statistical process control used in manufacturing nationwide since the 1950sSix basic rules in the Westgard scheme: 1-3s, 2-2s, R-4s, 1-2s, 4-1s, and 10x. These rules are used individually or in combination (multi-rule) to evaluate the quality of analytical runs.Detect random or systematic (shifts or trends) errorsWarning 12SD or 1-2s:It is violated if the single IQCvalue exceeds the mean by 2SD.
32 (mean + 2s or mean - 2s) limit. Rejection 22SD or 2-2s:This rule detects systematic error and is applied within and across runs.It is violated within the run when two consecutive control values exceed the "same" (mean + 2s or mean - 2s) limit.The rule is violated across runs when the previous value for a particular control level exceeds the "same"(mean + 2s or mean - 2s) limit.Within run violationAcross run violation
33 Rejection 13SD or 1-3s: Rejection 41SD or 4-1s: It is violated when the single IQCvalue exceeds the mean by 3SD.This rule is applied within controlmaterial only.The 1-3s rule identifies unacceptablerandom error or possibly thebeginning of a large systematic error.Rejection 41SD or 4-1s:It is violated if four consecutive IQCvalues exceed the same mean plus 1sor the same mean minus 1s control limit.
34 Rejection 10x:This rule detects systematic bias and is applied both within and across control materials.It is violated across control materials if the last 10 consecutive values, regardless of control level, are on the same side of the mean.The rule is violated within the control materials if the last 10 values for the same control level are on the same side of the mean.
39 Multi Control QC RulesThe multi control QC rules are followed in the department as described below:The rules to follow when one level QC material is used:Reject QC if:i) it is outside 3 SD (13s)ii) two consecutive values obtained are outside 2 SD on the same side but within 3 SD (22s)iii) ten consecutive values are above or below the mean, but within 2 SD (10x)The rules to follow when 2 level QC materials are used:i) either QC values is outside 3 SD (13s)ii) both QC values are outside 2 SD on the same side, but within 3 SD (22s)iii) difference between both QC values is >4 SD i.e. one level QC is > 2 SD and other level QC is <2SD (R4s).iv) ten consecutive values of the same level QC are >/< the mean, but within 2 SD (10x).v) five consecutive values of one level QC and five consecutive values of other level QC are >/< the mean but within 2 SD (10x)
41 Systematic vs. Random Errors Systematic ErrorAvoidable error due to controllable variables in a measurement.Random ErrorsUnavoidable errors that are always present in any measurement. Impossible to eliminate
42 Review to Manage “Out of Control” Situation: Random Errors: Search for recent events/changes:New reagent kit or lotNew control bottleInstrumentation component replacementInstrument maintenanceInstrument moveSearch for sources of Random Error:Power supplyDouble pipetting of control sampleMisplacement of control sample within the runAir bubbles in water supplyRandom air bubbles in reagent or sample pipette systemIncorrect reconstitution of the control productInappropriate storage of control in frost free freezersUse of non-reagent grade water in the test systemOperator technique
43 Corrective Actions Sources of Systematic Errors: Improper alignment of sample or reagent pipettesDrift or Shift in incubator chamber temperatureInappropriate temperature / humidity levels in the testing areaChange of reagent or calibrator lotDeterioration of reagent while in use, storage or shipmentDeterioration of calibrator or control product while in use, storage or shipment.Incorrect handling of control product (e.g. Freezing when not recommended)Inappropriate storage of control products in frost free freezersFailing light sourceUse of non-reagent grade water in the test systemRecent calibrationChange in test operatorSpecimen carry-overObstruction of tubingCorrective ActionsClassify error, inform n documentReview error for cause – detailedRelate error to causeRerun QC/retained sampleFollow manufacturer’ troubleshooting guideRun fresh controls if reqdCall for application’s support if…Take appropriate CAPAs actions and document the details in relevant “QC Failure Log”
44 External Performance Testing Blind samples submitted to laboratoriesLabs must periodically analyze in order to assure Quality for acceptable resultsAccreditation Reqts – Mandatory Cl 5.6???
45 Proficiency Testing/EQAS Process the PT samples in the same run as patient samplesProcess as per the instructions given along with the PT sample & not in duplicateDon’t send for analysis to another or referral laboratoryDo n’t discuss the results with another laboratory till the evaluation reports from the PT provider are availableEVALUATION OF PT PERFORMANCE:Review Unsatisfactory PT result for remedial corrective actions to be taken. See if there are any trends or shifts.Review Tips : Check forClerical errors & operator investigationIQC data for the date PT samples were run, stability, expiry, peer group Z score, trends ans shifts etcReagent logs – expiry, onboard stability, contamination, suitable calibration for new lotCalibration logs – cal curve, schedule to determine how close to the PT samples the calibration was performedEquipment logs- maintenance performedRerun frozen PT sample if availableOnce the problem that caused the PT failure is found, it is rectified, documented and reviewed with laboratory staffAll un-graded PT surveys are investigated to the acceptability of its performance with the same rigor as if it were an unacceptable performance followed by corrective action and documentation if requiredAll the participant statistics are reviewed and documented
46 Qa --- Other practices in lab QC Lot to Lot validation: Establishing Lab Reference Ranges with change in QC LotAssayed QC: Compare for two days old lot with new lotUnassayed QC: Do as new till then data points are evaluatedNew Test ValidationValidation of Calculated ParameterProficiency Testing/EQASCalibration:As indicated by calibration frequencyWhen Quality Control is outside RangeNew lot of Reagent IntroducedNew lot of Control is usedNew company QC reagent is usedReagent/Kit validationWhenever reagent/kit, Kit lot no changes, run controls and/or patient samples which have been previously tested by the old reagent/kit to detect any significant changesAccept if within 2SD. If unacceptable, then repeat in duplicate and record SD/CV